The Cenozoic strontium isotope (87Sr/86Sr) variation in seawater is largely defined using Deep Sea Drilling Project (DSDP) samples. Examination of non-DSDP results, particularly for the Paleogene, suggest that the original seawater curve of Burke et al. (1982) may lie above the true seawater ratio during this time. To investigate this possibility, we analyzed the 87Sr/86Sr for 230 Paleogene rock and shell samples from the 83 sites in Alabama, Mississippi, and Louisiana. Many of the samples are from classic sections on which the Gulf Coast Paleogene stratigraphy was originally defined. Multiple analyses of samples from these units show very little scatter, and in most cases, precise agreement is obtained among geographically separated collection sites. We believe that analysis of multiple samples from a unit of established age, collected at more than one site, yields results that most accurately define the actual seawater Sr isotope ratio at any given time. Results from these units define a seawater variation line that tracks well within the range of DSDP data, but is generally below the original seawater line of Burke et al. (1982). It is often assumed that samples from continental settings will exchange with high-ratio Sr thereby raising 87Sr/86Sr values. Our Gulf Coast samples, which are derived from a continental setting, yield individual 87Sr/86Sr values that are both higher and lower than the estimated contemporaneous seawater ratio. An analysis of our Gulf Coast data and DSDP results show that the direction of isotopic exchange cannot be confidently predicted for individual samples based solely on their derivation from either continental or oceanic settings. ¿ American Geophysical Union 1993